Performance Of Nature Inspired Routing Algorithms Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

And here this research will give information about different nature inspired routing algorithms like AntNet and Beehive performance by comparing them with packet delay, throughput etc in different traffic models. And also we can know which routing protocol we will use in wired and wireless networks. [1]

The state of the art nature inspired routing algorithms are AntNet, Beehive, DGA (Distributed Genetic algorithm). [2]

These routing algorithms were inspired by nature insects like ant, bee, etc This telecommunication networks were became a focus on research on academia and industry because there was a unexpected growth in internet users as it is created in nerve center of the infrastructure. And the reason of success in internet is there is no connection between sender and receiver (connectionless packet switching technology). [3]

In the past few years the researchers were in the field of natural computing and they have changed interest in creating a nature inspired routing algorithms. Here the routing algorithms are now implemented during the information limited in the routing table separately existing at every node of the network, and we know that traditional routing algorithms don't have sufficient flexibility to assure new routing needs and also they need a human support in order to adjust themselves changes and failure.

At the end of the Masters project, the research would show that how nature inspired routing algorithms will perform in different traffic models and different network conditions. [4]

2. Literature Review:

Routing is a process of choosing a path in the network from source to destination a packet is travelled. In the communication network routing plays a decisive role to determine the performance of network in criteria of throughput, packet delay.

Here the routing algorithms are now implemented during the information limited in the routing table separately existing at every node of the network, and we know that traditional routing algorithms don't have sufficient flexibility to assure new routing needs and also they need a human support in order to adjust themselves changes and failure.

Routing algorithm:

Routing algorithm is define has the processing inside router that will tells the information about routing. The routing algorithm main theme is to tell the best path and quality of service. And also the routing algorithm will detect the information if there is any change of the network or disconnect of links. And the routing algorithm will change with different routing protocols. [5]


Nature is a source of inspiration for creating a meta heuristics to sort out with different problems. And meta heuristics is not a single problem but we have framework for solving class of optimization trouble. McCulloch and Pitts began to form most direct mappings of biological studying process to computer algorithm in 1943. And this section is called as ANNs (Artificial Neural Networks) and this is based on neurology.

In 1985 Hopfield and tank, ANNs has some applications like combinatorial optimization and this was started by the scheme of Hopfield network for the "Traveling salesman problem" (TSP).

After that in 1975 another meta heuristic algorithms were inspired by nature and after that this were researched by (Holland) are called as EAs (Evolutionary Algorithms).

And the main legislature of computational paradigm were Gas (Genetic Algorithms), EP (Evolutionary programming) and ESs ( Evolution Strategies) are developed by Thomas in1996.

And the last fifteen years the combinatorial optimization problems were leading edge of heuristic. And after that a new field were created in terms of application is Swarm Intelligence (SI) by Beni in 1988, Beni and Wang in 1989 to tell the robots coordination. After that time there were 2 main solutions of Swarm intelligence for solving optimization problems.

One was PSO (Particle Swarm Optimization) which is addition technique inspired by social behaviour of fish schools or bird flocks which have been proposed by Kennedy in 2001.

And the other hand we have Swarm intelligence is defined as any plan which employs the swarm mechanics and which are derived of social insects for solving optimization problems Bonabeau in 1999. [6]

Figure: Routing protocols for fixed telecommunication networks [7]

Routing algorithms are classified by the developers has design philosophy. All the researchers in each and every community have taught with certain design and the analysis doctrine, they leave some space for cross fertilization of ideas like different communities. The reader can understand the principles of the different types of routing algorithms. They can develop state of the art routing algorithms for the network by these communities by knowing the various background.

The communities are discussed in the sequel:

Network community has initiated the work in the area of the packetswitched networks and the basic of the work was developed from ARPANET and novel routing algorithm and this was based on Bellman and Ford algorithm. After that they were many routing algorithms were created like dynamic and multi path routing algorithms this are based on classic methodology for creating routing protocol. They were link state packets which are non intelligent are used to collect the information about the neighbours costs and then circulate the information to other network. After that many change were created that means in order to get the shortest path by which it travel in the network and it requires a global system like Dijkstra's shortest path algorithm.

Here we have other community which is Artificial Intelligence routing community which will work in 2 different ways are Agent based learning and Machine learning. The machine learning is uses reinforcement learning to get the plan routing algorithms for packet switched networks and which are decentralised, deterministic, adaptive, dynamic and local. For this examples are like PQ routing and Q routing.

Here the exact routing algorithms are agent based learning methods.

And the advantages of these algorithms are

Here the management of network can gets a kind of benefit by using mobile agents.

Priori global system of this network does not require for this algorithm.

In the network agents decentralised and in the visited nodes they leave information. By this there is no need of global controller. And these algorithms can handle the changes in traffic patterns or network.

The main theme of this routing algorithms is to design intelligent agents for management, control of network and routing in the autonomous manner.

Here the 2 major direction of Natural computing research are evolutionary computing and swarm intelligence.

And the evolutionary computing will takes growth in the process of the living cells. And this will develops evolutionary operations like mutation, croddover and selection. And here Distributed genetic algorithm (DGA) is one of te routing algorithms.

And the swarm intelligence is like it acts as self process in nature and this are utilizes the principles as the moving metaphor to get novel solutions for different scientific problems. In this population system has local knowledge but they form together to get an intelligent system. Here the ABC, AntNet and Beehive are belongs to this routing algorithms. And nature inspired routing algorithms are local, probabilistic, adaptive, dynamic and decentralised.


Ants in real life:

Being a individual ants are simple insects but if they are part of colony they can do any complex tasks easily. In real life the ant behaves like: they will build and protect their nests, for carrying large amount food they will cooperate, colony immigrate, and they have limited memory, for food to nest they will find shortest path, etc[9]


The social insects can communicate in the form of stigmergy which are like ants. Here the ants are travelling from nest to food source by contribution information from other ants through stigmergy and also to find the shortest path.

Figure : Ants laying pheromones[10]

Ant colony optimisation:

Here the first ACO (Ant colony optimisation) meta heuristic was inspired by ants operating principles, the ants creates a colony like nest building and foraging which is complex task. In this we have two state of the art routing algorithms were created like ABC, Antnet. The ABC algorithms were designed on the basis of circuit switched telecommunication network and Antnet algorithm is desined on the basis of packet switched telecommunication network.

Ant based control (ABC):

Here we different forms of ant based control algorithm which are Ant based control for circuit switched networks and Ant based control for packet switched networks.

ABC (Ant Based Control) for circuit switched networks:

Schoonderwoerd was first apply from ACO met heuristic routing and load balancing problems in the circuit switched telecommunication networks.

ABC (Ant Based Control) for packet switched networks:

Subramanian was first created Ant based control for packet switched networks. And they are creates 2 types of ant like uniform and regular.


Di caro and Dorigo was proposed AntNet. And this is inspired from principles of ACO metaheuristic and other additional network conditions. And this is created as asymmetric packet switched network, to get maximium performance from the network. By this now we are getting a great success by solving the practical and theoretical optimisation.

In this network two agents were monitored: forward Ant and backward ant. The forward ant agent is created at normal intervals from source to assured destination. And this ant agent will equipped with a stack memory on which entrance time and address of each node is pushed to that path. And after that if the forward ant reaches the destination it will creates a backward ant agent and backward ant will take all information from forward ant. And the backward ant will go in the same path that forward ant is travelled and it will update the time.

And it has features like robustness, decentralised in nature and adaptation. And in modern network we are routing network as mobile ad hoc network (MANETs) which is basis on ant algorithms and also the queuing network analysis.[11][12]

Honey Bee in real life:

The bee colony looks like nature and each bee first looks for food alone. When the bee finds the food that will dance to tell other bee members that there is a food. And the entire bee will collect the food to hive. After taking the food to the hive, the bees can gives 3 different ways.

Dump the before food source and become another time uninterested admirer.

Without tell to nestmates continue the forage at food source.

Dance and tell the nestmates before coming from food source. [13]

BeeHive algorithm:

After the implementation of ant routing algorithms which were successful. But in 2004, nobody has studied the high efficient resource like the pattern of honey bee colonies which is in the field of information technologies and this is known as nature inspired routing algorithms.

Figure: Honeybee [14]

And the Beehive algorithm is created by Weede, Farooq and Zhang in and this was inspired by honey bees communication language. And the beehive algorithm is developed a flexible, this will gives high throughput and fault tolerance and highly adaptable. And after that they effectively modified that algorithm to BeeAdHoc (Mobile Ad Hoc network) and for traffic control a BeeJamA(prevention of traffic in road networks) network is designed.

Here the each node sends an agents in regular intervals to its neighbour sites. It uses prority queues. In this algorithm it utilizes only forward moving agents and it oppose to Antnet in this no arithmetic parameters are stored in routing tables. In the Beehive network it is divided into foraging zones nad foraging region. Here each node is delegate to one foraging region and each foraging region has a node. It is also a fault tolerant to crashing of the routers. Here the next hop at a node is particular in probabilistic way depending upon goodness. [15]


The Beehive is inspired by foraging principles of honey bees and this intelligent bee agent is suitable for complex and large networks. By the extensive simulation experiments the result says that the bee agents require smaller band width and processor time is less. As compared with state of the art like AntNet routing algorithms the beehive gives simple and better performance.


It gives a solution for mobile network community from core dilemmatas. And this algorithm is simple and it requires 2 types of agents, they are foragers and scouts. Here the BeeAdHoc is classified as reactive source. This foragers are created carefully manner that this would packed into header of IP packet.


It also known as Bee inspired traffic jam avoidance. It is created because of traffic congestions problems in metropolitan areas, cities due to highly character of building and metals. And it is integrates with Beehive routing algorithm which give flexibility, high throughput, fault tolerance, high adaptability. [16]

Previous research on a related topic:

Previous research on this topic is "A comprehensive survey of nature inspired routing protocols" (by H.F.Wedde and M.Farooq in 2006) in this article they have briefly explained about routing protocols which are inspired by nature and this article says that they wants to motivate the researchers to develop state of the art routing algorithms.[17]

And other research on this topic is "A Performance evaluation framework for nature inspired routing algorithms" (by Horst.Wedde and Muddassar Farooq) this article tells that the performance of the nature inspired routing algorithms is compared with OSPF under static conditions. And also they have tested for performance framework for quality of service parameters. Here they have taken a simulation tool which is OMNet++. [18]


The purpose of this research is to show that we can eliminate the network wired, wireless devices were smaller when compare with wired devices and more affordable. By using this routing algorithm wireless link can be flexible and the speed of the internet will increases without any problems like quality of service. And by using this routing algorithm, the packets can travel in shortest path and accurate.

3. Methodology:

Here the method used to check the performance of routing algorithms by simulation tools like NS2. This can run only in linux. And the output of the system can check in NS2 like graphical views. Here the performance of the network is seen in NS2 output graphical representation. Here I need to know whether traffic network work in NS2 or not. Before that I need to learn the coding to know how it will work. Here I need to select a wired or wireless network condition like using MANET, Traditional routing.

First I need any operating system Linux and then I need to learn how the commands will work. After that install a simulation tool NS2 in the linuk operating system and then I need to learn how to use that commands in the NS2 simulator.

And I need to create a traffic network design and implement in the simulator. And then I need to give some routing algorithm in C++ and OTcl coding. I need to get from coding files of AntNet and Beehive routing algorithms.

After that I need to implement that code in that traffic model and run the simulation. And test for each interval of time and take different traffic model.And finally know the performance of different nature inspired routing algorithms, by knowing the value of throughput, delay and adaptability. And then compare the results and explain the results of both.

Simulation environment:

Here we have different types of simulations like NS2, Opnet, OMNet++ etc for calculating some network performance etc

Network simulator like NS-2 is a generally known simulation used in study of number of network algorithms and protocols using different scenarios. In wired and wireless networks the simulation like NS-2 supports. [19]

And here to make use of network simulator ns-2 for the implementation and testing approach for the existing method. Packetlevel simulator is NS-2. In 1987 Ns-2 is appeared as REAL and it is object oriented software package. And this is used in VINT project, LBL, Xerox PARC and ISI. And it is discrete event that means it schedules events aligned with time and in this events are observed one by one. The NS-2 coding is written in two programming languages like OTCL (Object oriented version of TCL) and C++. And this ns-2 project uses other tools like tracing, controlling the results of simulations and logging.

Here am selecting this NS2 because it is open source and it can find bugs in it. And it is very flexible and it returns the new developments of new technologies in a quick way than other network simulator.

But it has some disadvantages like lack of documentations and systematic and the result is not accurate compare to real networks.

Here am selecting NS-2 because it is a golden choice in research [20]


Throughput is nothing but a actual bit rate transfer to reach the destination from the route or link. It means that data that transferred from one place to another in a period of time. [21]

Network specification tools:

The network is designed as traffic models are:

Simulation tool NS2, OTcl

Linux operating system

Routing protocols like AntNet, BeeHive.

Library Files of NS2, OTcl

4. Limitations of the research:

The performance of the routing algorithms need to be checked but in different scenario. We need to take different traffic model and in different network conditions. In this research the network is limited by the amount of traffic. Here am using NS2 simulation tool and need to get results. When comparing with real network result they will not be same, because of increase in network traffic or else increase of applications. In real case scenario the different traffic models and changing network conditions may change with research scenario. And the result is not equal for both the conditions because of network equipments. In real life the types of equipments may change with the clients. And this research can only be done in few simulation tools, only with compatible operating systems. Here the coding can be done in C++ and OTcl. But it limited in real scenario because wired and wireless networks are complicated.

5. Expected results:

The expected results of this project are shown by graphical output of the network like throughput, delay, adaptability in different traffic models. But I need to check two routing algorithm performance like AntNet and Beehive routing algorithms. In that I need verify which routing algorithm gives good performance. By doing this we can know which algorithms can implement in future work and to get more accurate output of the network.

I expect that the Beehive performance is better than AntNet routing algorithm.

6. Project management:

Here am using waterfall lifecycle for my project. The waterfall lifecycle is correct selection for this project because it does not include any development and it is straight forward. But here I need to practise NS2 and need to learn basic of linux. This project will complete for 20weeks and this includes writing of the project proposal. And the project has started on 15th December 2009, and it will complete till 30th April 2010.

Proposed chapters for the main dissertation:




Literature Review

Network configuration



Results analysis


Future work

Bibliography and Appendixes

7. Gantt chart: